Calculations

The primary calculation used in High Bill Alerts (non-AMI) is the forecast calculation, the result of which is displayed in the Forecast module. The calculation works by determining how sensitive a customer's home is to hot and cold weather. Three major steps are involved in the calculation.

Step 1: Determine the heating, cooling, and baseload coefficients.

  1. Get historical weather data and historical energy use data for a customer. Two years of each type of data is strongly preferred. A minimum of 13 months is required.
  2. Perform a linear regression analysis on the data to see how a customer's energy consumption fluctuates in relation to changes in the weather. The goal of the regression analysis is to generate heating, cooling, and baseload coefficients.
    • A baseline temperature of 65 degrees Fahrenheit is used for the analysis. The baseline temperature is the temperature of the building at which no additional heating or cooling is required. At 65 degrees or less, most buildings require heat to maintain a 70-degree temperature inside. At 65 degrees or more, most buildings require air conditioning to maintain a 70-degree temperature inside.
    • We exclude anomalies from the data model such as average daily electric usage less than 1 kWh or greater than 300 kWh.
  3. Based on the results of the analysis, produce the heating, cooling, and baseload coefficients.
    • Cooling degree day (CDD) coefficient: The number of degrees above the baseline temperature for a period of time. For example, if the baseline is 65 degrees and the weather is 75 degrees, then you have 10 CDD.
    • Heating degree day (HDD) coefficient: The number of degrees below the baseline temperature for a period of time. For example, if the baseline is 65 degrees, and the weather is 60 degrees, then you have 5 HDD.

      Note: To calculate the heating and cooling coefficients, we select a model that is specific to your climate zone. That way, the forecast estimates are more accurate and more relevant to customers' local weather patterns. Our Data Science team has created different calculation models for regions with different climates. The models we choose from include:

      - 3-Parameter Heating (3PH): Used for extreme heating climates.

      - 3-Parameter Cooling (3PC): Used for extreme cooling climates.

      - 3-Parameter Heating-Cooling (3PHC): Used for seasonal climates.

    • Baseload coefficient: The kWh/day that a home uses minus the consumption due to heating or cooling.

Step 2: Estimate period-to-date energy usage.

  1. Get weather data for the bill period to date.

  2. Calculate the number of HDD in the bill period to date, and multiply the result by the heating coefficient produced in the Determine coefficients step above. Then do the same thing for CDD, and multiply the result by the cooling coefficient.

    For example, let's say that 20 days have passed by in the bill period to date. The calculation looks at each day to count the number of HDD or CDD, and then sums them up. In a winter month, for example, one day might have 5 HDD, another day might have 4 HDD, and so on. The calculation would add the number of HDD for all 20 days, and then multiply the result by the heating coefficient produced in the Determine coefficients step above.

  3. Calculate the number of baseline days (that is, how many days have passed since the beginning of the current bill period), and multiply them by the baseload coefficient produced in the Determine coefficients step above.
  4. Add the results of the coefficients together to get an estimate of period-to-date usage. For example, let’s say that in the first 20 days, we estimate that the household has used 760 kWh.

Step 3: Project the customer's usage for the rest of the billing period and trigger an alert if necessary.

  1. Divide the usage-to-date by the number of days in the bill period that have already passed. For example, if 20 days have already passed, we would divide 760 kWh by 20, resulting in a daily average of about 38 kWh per day.
  2. Multiply the result by the expected number of days in the bill period. For example, for the month of April, we would multiple 38 kWh * 30 to get a projected usage of 1,140 kWh.
  3. Compare the forecasted usage to the total usage of the previous bill period. The previous bill period may be the previous month or the same month from the previous year.
  4. Trigger the high bill alert if the forecast is 30% higher than the previous bill period.